Re: Notes on spec

/ Richard Tobin <richard@inf.ed.ac.uk> was heard to say:
| 2
|
| As well as saying that implementations must not assume that components
| are functional, we should explicitly say that they must not assume that
| they are side-effect free.

Done.

| Does a subpipeline have to be connected?  A pipeline doesn't.

I removed "connected". There's a larger question of whether
"subpipeline" is a good name for the set of steps inside a step
container.

| 2.1
|
| A component *has* a number of ports, which may be zero.  But it *MAY
| have* an arbitrary number of parameters.  Surely it *has* a number of
| parameters, which may be zero.

Fixed.

| 2.3
|
| Did we agree that the string value of non-string parameters is used?
| I remember arguing that it should be an error to use non-strings,
| so that a future version could compatibly allow non-string values,
| but I don't remember what we agreed.  Thinking about it now, perhaps
| it's not a problem since we could require non-string parameters to
| be declared as such.

I think we agreed that they're converted to strings.

| 3.5
|
| "guarantee" not "gaurantee"; "available" not "avialable".

Sigh. "Gaurantee" is just a word I cannot spell correctly.

Fixed.

                                        Be seeing you,
                                          norm

-- 
Norman Walsh
XML Standards Architect
Sun Microsystems, Inc.